• Apex.AI

Simplifying development of automotive production software with Apex.OS 2.0 & Apex.Middleware 1.0

Updated: Jun 3

Our customers have been busy developing their cars, trucks, robotaxis, agricultural tractors, and ADAS/AD offerings with Apex.AI supporting them to speed up the delivery of new vehicles. The insights and feedback from our customers have helped the Apex.AI products rapidly mature. These latest releases of Apex.AI products continue our practice of delivering innovations, enabling easier development, and providing better performance with each release.


We are excited to announce the releases of Apex.OS 2.0 and Apex.Middleware 1.0. In the runup to the release, we gave early access engineering releases to customers. Customers gave feedback that enabled us to refine the developer experience and improve features. These releases contribute to customer success with:


  • Ease of use with connectors for vehicle protocols and cloud that make it easy to use Apex.OS, ROS, SOME/IP, and CAN together in one vehicle. We also improved the developer experience for CLion and Visual Studio Code users.

  • Performance via DDS with zero-copy improvements and several other convenient optimizations that each give 20% latency improvement for your applications.

  • Safety improvements made for our OEM and Tier 1 customers that are targeting ASIL D in-context safety certification.

  • Save time by speeding development, getting more done with fewer people, and eliminating cross-compilation steps in your CI build farm.

  • Reference implementations providing practical and comprehensive examples of best practices.


Apex.AI included a new executor in a previous release. Now Apex.OS has migrated to this new execution paradigm making it the default. So these gains in performance and efficiency are what customers get straight out of the box along with simulation time, deterministic record & replay, and APIs to manage time concepts.


Apex.Middleware 1.0 incorporates the Apex.AI co-developed Eclipse iceoryx 2.0, which is a functionally complete inter-process communications (IPC) mechanism for AUTOSAR ara::com, Apex.OS, and customer frameworks. Apex.Middleware now has CommonAPI + SOME/IP binding and tools for Adaptive AUTOSAR compatibility. It has SocketCAN support built-in.


Apex.Middleware 1.0 provides an additional 20% reduction in latency for DDS with zero-copy compared to the previous Apex.Middleware 0.9 release. For example, 48 µs pub/sub for 4MB camera images on a 4-core Arm® Cortex®-A53 Renesas R-Car VH3 with QNX® OS for Safety 2.2. And this release puts into place what we need to safety certify Apex.Middleware.


Other convenient optimizations that are now made easier to use include executing multiple nodes in a single thread, avoiding context switches in the middleware, and identifying non-triggering topics. Each of these provides 20% latency improvement and CPU time reduction in our testing of an ADAS application. The blog A Journey from Microcontrollers to Microprocessors with AUTOSAR & ROS explores this in more detail.


With these new features, developers have any-to-any communications for applications they create to work with their vehicle’s other applications, automotive components, and cloud. Apex.AI is working to combine the best of ROS, AUTOSAR, and other frameworks to enable developers to cover the most use cases while avoiding being locked into a single communication mechanism.






Our customers are central to how Apex.AI defines, designs, and develops products. So these Apex.OS and Apex.Middleware releases deliver what our customers need:

  • Customers gained support for Arm native parity end-to-end to develop on MacBook Pro M1 with CI/CD DevOps on AWS Graviton2 and Ampere Altra. They also received Qualcomm Snapdragon platform support and further improved AWS IoT support.

  • Another customer now is using a production platform based on NVIDIA AGX Xavier SoC, and the Apex.OS system monitor is integrated into their safety concept to trigger the safety mechanism on the ASIL-D microcontroller for safe recovery. They also now have CAN support for their Tier 1’s safety cores.

  • For a different customer we realized CAN support for their ADAS products, the flexibility for large complex systems with hundreds of subscribers, and support for their new compute platform.

  • A customer now has a developer onboarding experience tailored to their needs including Visual Studio Code with Apex.OS tutorial, fresh documentation for their key use cases, and lots of customized training sessions.

Since the previous release, many more amazing engineers have joined Apex.AI. To give great support to our growing user base, Apex.AI has doubled the size of the Application Engineering team supporting our customers. We are hiring. We are also building out the Apex.AI Solution Center in Palo Alto, so contact us to arrange a visit. See you here!


If you are interested in Apex.AI products for your projects, Contact Us.

We are also growing our team worldwide. Remote and in our Labs. Visit Careers.





1,274 views

Recent Posts

See All